© 2008 Nature Publishing Group http://www.nature.com/naturechemicalbiology
news and views
dynamic manner, may therefore restore normal
tau glycosylation and thus prevent pathological
phosphorylation.
The design of the O-GlcNAcase inhibitor
Thiamet-G (2a), as Yuzwa et al.2 have termed
their agent, represents an elegant example of
structure-based drug design. Stabilization of
high-energy transition states that occur during the chemical conversion of substrates to
products is at the heart of enzyme catalysis.
Thiamet-G is a stable compound whose fused
thiazoline ring system (red in 2) geometrically
mimics a transition state of the substrateassisted enzymatic hydrolysis of protein-OGlcNAc units and, in this way, effectively inhibits
O-GlcNAcase function. Similar oxazoline systems that inspired the design of Thiamet-G are
found in natural products with glycoside hydrolase inhibitory activities, such as allosamidin (3)
and allosamizoline (4)3, as well as in synthetic
derivatives (such as 2b and 2c)4. Unlike known
O-GlcNAcase inhibitors, Thiamet-G, designed
by reference to X-ray crystallographic structures
of hexosaminidase–inhibitor complexes, shows
high selectivity for O-GlcNAcase over related
hexosaminidases, especially lysosomal Hex A
and Hex B. For example, replacement of simple
small thiazoline 2′-substituents with the ethylamino group of Thiamet-G increases not only
affinity by virtue of favorable electrostatic interactions between the inhibitor amino group and
the enzyme’s aspartatecatalytic residues (Fig.
1b–d) but also enhances selectivity through
optimal occupation of the pocket that accommodates the substrate N-acetyl group, which is
noticeably larger in O-GlcNAcase than in Hex A
and Hex B (Fig. 1e–g). Because deficiency of Hex
A and Hex B is associated with several serious
lysosomal storage disorders (such as Tay-Sachs
and Sandhoff diseases), inhibitor selectivity for
O-GlcNAcase over lysosomal hexosaminidases
is important for precisely elucidating the roles
of these proteins, as well as for potential therapeutic safety in drug development.
Apart from its enzyme potency and selectivity, the notable features of Thiamet-G as a
first-generation pharmacological O-GlcNAcase
inhibitor are its oral bioavailability and ability
to cross the blood-brain barrier. According to
Yuzwa et al.2 the secondary amino functional
group of Thiamet-G shows a pKa value of 8.0,
which is likely to be beneficial not only for
enzyme affinity but also for CNS bioavailability,
as compounds that are somewhat basic under
physiological conditions are more likely to be
blood-brain barrier permeant5. In neuronal
cell culture, Thiamet-G was able to reduce tau
phosphorylation at pathologically relevant sites
up to threefold. Immunohistochemical analysis
of brain samples from animals that had received
Thiamet-G by the oral administration route
revealed a similar picture of specific suppression of tau phosphorylation.
Recent results suggest that direct inhibition
of the kinases that are implicated in pathological tau phosphorylation may be a viable
therapeutic strategy for tauopathies (reviewed
in ref. 6). The results of Yuzwa et al. now suggest an alternative and indirect strategy. As we
saw, decreased glucose availability in the brain
limits O-GlcNAcylation and thus allows tau
hyperphosphorylation. Elsewhere in the body,
however, excess glucose can lead to an increased
flux through the hexosamine biosynthetic pathway, leading to elevated O-GlcNAcylation of
many proteins, some of which are known to
contribute to metabolic disease states such as
hyperglycemia, hyperinsulinemia and hyperlipidemia, as well as the adverse affects of
these disorders on the heart7. For this reason,
it might be expected that global inhibition of
O-GlcNAcase could result in undesirable side
effects. Although Yuzwa et al. do not report any
gross toxicity of their O-GlcNAcase inhibitor
upon in vivo administration, even at what seems
to be a very high dose, further O-GlcNAcase target validation and Thiamet-G toxicology studies
are clearly warranted. Apart from representing
an interesting new drug lead, Thiamet-G will
certainly be a valuable tool for the in vivo study
of the role of O-GlcNAcase in tauopathies.
1. Hart, G.W., Housley, M.P. & Slawson, C. Nature 446,
1017–1022 (2007).
2. Yuzwa, S.A. et al. Nat. Chem. Biol. 4, 483–490
(2008).
3. Sakuda, S., Isogai, A., Matsumoto, S., Suzuki, A. &
Koseki, K. Tetrahedr. Lett. 27, 2475–2478 (1986).
4. Knapp, S., Kirk, B.A., Vocadlo, D. & Withers, S.G. Synlett
435–436 (1997).
5. Lipinski, C.A. Annu. Rep. Comput. Chem. 1, 155–168
(2005).
6. Mazanetz, M.P. & Fischer, P.M. Nat. Rev. Drug Discov.
6, 464–479 (2007).
7. Fulop, N., Marchase, R.B. & Chatham, J.C. Cardiovasc.
Res. 73, 288–297 (2007).
8. Mark, B.L. et al. J. Mol. Biol. 327, 1093–1109
(2003).
9. Dennis, R.J. et al. Nat. Struct. Mol. Biol. 13, 365–371
(2006).
10.Lemieux, M.J. et al. J. Mol. Biol. 359, 913–929
(2006).
11.Whitworth, G.E. et al. J. Am. Chem. Soc. 129, 635–644
(2007).
Protein dynamics under light control
Michele Vendruscolo
A stochastic view of allostery is providing quantitative estimates of the energy made available through protein
photoswitches.
Living cells constantly monitor their own state
and their surroundings in order to respond
effectively to changes in them. Their ability to
perform these tasks relies on the presence of
complex networks of interacting proteins that
sense the environmental conditions and enable
appropriate biochemical reactions to take place
to maintain homeostasis and promote development. At the molecular level, the transfer
Michele Vendruscolo is in the Department of
Chemistry, University of Cambridge, Lensfield
Road, Cambridge CB2 1EW, UK.
e-mail: [email protected]
of information is often achieved through the
relay of highly specific conformational changes
within and between proteins. A longstanding
challenge has been to understand how much
energy is required to initiate and sustain these
processes. As reported in this issue, Yao, Rosen
and Gardner have now been able to provide a
quantitative estimate of the energy made available by a protein photosensor1.
Key to the result obtained by Gardner and
co-workers1 has been the recognition that functionally important conformational changes of
proteins are often achieved through a shift in
the equilibrium populations of interconverting
nature chemical biology volume 4 number 8 august 2008
states. According to this view, conformational
fluctuations of proteins, rather than being uniformly distributed in a random manner, are
organized to prompt them to sample their active
states2,3. This dynamical view of protein behavior, which has already generated major advances
in understanding enzymatic catalysis4,5 and
allosteric activity6, has now been applied to the
study of sensory proteins.
To study the molecular basis of the photosensor activity of proteins, Gardner and co-workers1
focused their attention on phototropins—bluelight receptor proteins that control the ability of
algae and plants to carry out photosynthesis by
449
© 2008 Nature Publishing Group http://www.nature.com/naturechemicalbiology
news and views
regulating a variety of processes, including phototropism, chloroplast movement and stomatal
opening7. As with other photosensory proteins,
phototropins function by binding light-absorbing chromophores to convert the energy of photons into structural and dynamical changes. The
absorption process itself takes place in two socalled LOV domains in the N-terminal region
of the phototropin. The subsequent conformational changes are then transmitted to the kinase
domain at the C-terminal region of the protein.
Previous studies identified as a key step in the
signaling process the partial unfolding upon
light absorption of Jα, an α-helical connective
element between the second light-absorbing
LOV domain (LOV2) and the kinase domains8.
The energy involved in such an event, however,
was not precisely known.
By adopting a stochastic view of protein
allostery, Gardner and co-workers have been
able to fill this gap, and to demonstrate that the
conformations dominant in the lit state of the
phototropin that they studied are present also
in the dark state, although with a low statistical weight1. They exploited the opportunities
offered by NMR spectroscopy to measure, at
least in favorable cases, the relative populations
of states and the rates of their interconversion
with high accuracy. By using the Carr-PurcellGill-Meiboom (CPGM) relaxation dispersion
technique9, which enables detailed information to be obtained about low populated states
in equilibrium with the native state, they first
estimated the free energy difference between the
ground and the excited states of the LOV2-Jα
complex in the dark as 2.4 kcal mol–1 (Fig. 1a).
Then, by comparing the chemical shifts of the
Jα peptide in isolation and in the lit state, they
estimated the free energy difference between the
ground and the excited states in the presence of
light as –1.4 kcal mol–1 (Fig. 1b). Hence, they
concluded that upon light absorption a total
of 3.8 kcal mol–1 becomes available for transmission from the LOV2-Jα complex to the
phototropin kinase. Thus, they have been able
to quantify the free energy associated with this
conformational change, and in the process to
reveal that this particular photoswitch functions
by changing the equilibrium populations of the
dark and the lit states.
As 3.8 kcal mol–1 is almost 20-fold smaller
than the energy made directly available by the
absorption of a blue-light photon1, phototropins would not seem to be very efficient as light
sensor devices. However, the energy that they
yield is close, for instance, to that provided
through ATP hydrolysis, which is itself comparable to the overall stability of most globular
proteins10. Thus, this amount of energy seems
to be of just about the right scale to prompt
functional conformational transitions in folded
450
a
LOV2
∆Gdark
LOV2
FMN
Jα
FMN
Jα
b
LOV2
FMN
Jα
∆Glight
LOV2
FMN
Jα
Figure 1 Equilibrium shift of a protein photoswitch. (a) In the absence of light, the LOV2-Jα complex
fluctuates on the micro- to millisecond timescale between the dark state and a lit-like state, with a free
energy difference ∆Gdark = 2.4 kcal mol–1 between them. (b) By contrast, in the presence of light, the
covalent binding of a chromophore (flavin mononucleotide FMN) to LOV2 triggers conformational and
dynamical changes leading to a reversal in the free energy difference, which becomes ∆Glight = –1.4 kcal
mol–1, between a dark-like state and the lit state. Hence, light absorption makes 3.8 kcal mol–1 available
to the protein for initiating its kinase activity.
proteins. Therefore, rather than being directly
related to the efficiency of molecular machines,
the specific amount of energy generated by
phototropins seems to be scaled down from the
substantial energy of photons to the right level
to alter specific conformational states of proteins. These results obtained by Gardner and
co-workers open the possibility of generating a
variety of photoswitches with a large dynamical
range by altering—for example, through protein
engineering10—the efficiency of the conversion
of the energy of photons into conformational
and dynamical changes of proteins.
The ability to make quantitative estimates
of the energy provided in proteins through
light absorption, or indeed through other sensory processes, holds the promise of providing powerful further tools to engineer protein
sensors with accurately tunable responses.
These advances will be firmly rooted in the
view that the control of protein behavior—
including their signaling activity—can be very
effectively achieved through the modulation
of their equilibrium structural fluctuations.
1. Yao, X., Rosen, M.K. & Gardner, K.H. Nat. Chem. Biol.
4, 491–497 (2008).
2. Mittermaier, A. & Kay, L.E. Science 312, 224–228
(2006).
3. Vendruscolo, M. & Dobson, C.M. Science 313, 1586–
1587 (2003).
4. Eisenmesser, E.Z. et al. Nature 438, 117–121 (2005).
5. Boehr, D.D. et al. Science 313, 1638–1642 (2006).
6. Volkman, B.F. et al. Science 291, 2429–2433
(2001).
7. Christie, J.M. Annu. Rev. Plant Biol. 58, 21–45
(2007).
8. Harper, S.M., Neil, L.C. & Gardner, K.H. Science 301,
1541–1544 (2003).
9. Palmer, A.G., Kroenke, C.D. & Loria, J.P. Methods
Enzymol. 339, 204–238 (2001).
10.Fersht, A.R. Structure and Mechanism in Protein Science
(Freeman, New York, 1999).
volume 4 number 8 august 2008 nature chemical biology